How Was Energy Usage Information Collected in the 2012 CBECS?
CBECS 2012 - Release date: March 18, 2016
The Commercial Buildings Energy Consumption Survey (CBECS) project cycle spans at least four years, beginning with development of the sample frame and survey questionnaire and ending with release of data to the public. This set of three methodology documents provides details about each of the three major stages of the 2012 CBECS survey process.
- • How Were Buildings Selected for the 2012 CBECS?
- • How Was the 2012 CBECS Buildings Survey Conducted?
- • How Was Energy Usage Information Collected in the 2012 CBECS?
Overview
Energy usage data collection is a two-phase process in the Commercial Buildings Energy Consumption Survey (CBECS). Both phases involve multiple levels of data checks to verify that the energy usage information is accurate and that it corresponds to the sampled, interviewed CBECS building, which is critical to the quality of the final estimates
Phase 1 is the Buildings Survey (CBECS), in which interviewers collect building characteristics and energy usage data (consumption and costs) from a respondent at the building. The buildings survey data collection took place between April and November 2013. If the building respondent does not provide acceptable energy usage data during the interview, they are asked to provide their energy supplier's name and account numbers and the building moves on to Phase 2.
Phase 2 is the Energy Supplier Survey (ESS), which is a follow-up survey of the energy providers for buildings without accurate energy data from the CBECS. EIA contacts providers of electricity, natural gas, fuel oil (note: the fuel oil designation includes distillate fuel oil, diesel, and kerosene), and district heat (steam or hot water) to collect the needed energy usage data. The energy data are collected using a secure website that offers several reporting options designed to minimize reporting burden. The ESS data collection took place between March and October 2014.
At the end of the two phases of energy data collection, each building case ultimately has an energy usage value for each energy source used. Those values can come from one of three sources: the CBECS building respondent, the ESS, or, if neither of the two survey phases yields a plausible value, from data imputation using an engineering-based regression model.
This document details each phase of the energy data collection for the 6,424 buildings in the final CBECS sample that are not strip shopping center buildings. Strip shopping centers are treated differently and information about that methodology will be published at a later time.
Phase 1: buildings survey
In the CBECS buildings survey, the interviewer asked the building respondent to provide the consumption and/or expenditures data for each energy source that is used in the building (electricity, natural gas, fuel oil, and/or district heat). The following table provides the number of buildings that were reported to use each energy source and the number of buildings for which respondents were able to provide either their consumption and/or expenditures (C/E) data during the interview.
Cases that reported using the source | Of these, cases where the CBECS respondent was able to provide consumption and/or expenditures (C/E) | |
---|---|---|
Electricity | 6,257 | 3,371 (54%) |
Natural gas | 4,172 | 2,052 (49%) |
Fuel oil | 1,522 | 840 (55%) |
District heat | 492 | 186 (38%) |
During the CBECS interview, the computerized survey instrument ran an algorithm that predicted how much energy the building might use based on its reported characteristics. When the reported electricity, natural gas, or fuel oil usage value was significantly more or less than the expected value, the survey instrument prompted the interviewer to ask if there was anything unusual that might explain the high or low consumption. Whether or not the respondent provided an explanation, they were prompted to provide their energy supplier information, and the case was automatically routed to the ESS. The ESS was also initiated if the respondent did not know their heating, cooling, or water heating energy sources, or if they reported that the energy use in more than 25% of the building’s square footage was excluded from the figure(s) provided. The numbers of cases where the respondent provided C/E data but the case was subsequently sent to the ESS are shown in the next table.
Cases where C/E were provided by the CBECS respondent | Of these, cases that failed checks and were automatically routed to the ESS | |
---|---|---|
Electricity | 3,371 | 206 (6%) |
Natural gas | 2,052 | 237 (12%) |
Fuel oil | 840 | 12 (1%) |
District heat | 186 | (*) | (*) The energy prediction algorithms were not used for district heat, so cases were not automatically routed to district heat suppliers unless the respondent could not provide their district heat usage. |
After the CBECS data were received by EIA, the respondent-provided data that passed the initial checks during the interview were subject to additional data checking that was more rigorous than the interview checks. CBECS editors analyzed energy intensities (consumption per square foot) by building activity and compared the values to those from the previous CBECS. The case was reviewed if the reported value was significantly different from the previous-CBECS intensity for a given building activity (note: for fuel oil, the comparison was not done by building activity, but instead by whether or not the building used fuel oil for space heating). Cases were also reviewed when the respondent did not provide data to cover all of 2012 (or the months that the building was used in 2012), when energy usage in 10% to 25% of the building’s square footage was excluded from the usage figures (if more than 25% of the square footage was excluded, the case automatically went to the ESS), when the respondent indicated either that part of the building was left out or that additional buildings were included in the figures but they didn’t know the square footage of those areas, or if the price fell out of expected bounds.
In all of these situations, EIA reviewed the case in detail, using resources such as scanned energy bills provided by the respondents, consumption and expenditure worksheets that respondents filled out prior to the interview, audio recordings of the interaction between the interviewer and the respondent, and Google Earth and other internet resources. For each case, EIA made a decision whether to keep the data as is, route it to the ESS, or impute it. For fuel oil and district heat, if the data were unusable, consumption and expenditure data were almost always imputed because determining the energy supplier for these sources based just on the location of the building would have been very difficult. The decision to impute the data rather than send the case to the ESS was also often made in campus situations where it was clear that the individual building was not separately metered and it was unlikely that the supplier would be able to provide data for the individual building.
The next table summarizes the outcomes of the cases where the energy data was originally provided by the CBECS respondent and not automatically routed to the ESS.
Cases where C/E provided by the CBECS respondent passed initial chec ks and were therefore not routed to the ESS | Of these, cases that were flagged for C/E review | C/E review decision = ESS | Cases reviewed for other reasons where decision = ESS | |
---|---|---|---|---|
Electricity | 3,165 | 346 | 175 | 52 |
Natural gas | 1,815 | 215 | 87 | 35 |
Fuel oil | 828 | 101 | 0 | 0 |
District heat | 186 | 14 | 2 | 0 |
Occasionally during editing, it became clear that the original reporting of an energy source used was incorrect, i.e., it was originally reported that the building didn’t use the energy source, but most likely it did, or vice versa, so, as reflected in the tables in this document, the number in the sample using the source changed somewhat from the start of collection to the end of data editing and imputation.
As part of CBECS data quality research, there were some cases that were routed to the ESS for research purposes even though the respondent provided acceptable electricity or natural gas consumption or expenditures data (i.e., the figures didn’t fail instrument checks or post-collection checks).
The next table shows the breakdown of cases at the end of Phase 1.
Cases that used the energy source | Cases with verified C/E data provided by CBECS respondent | Cases with no C/E data or questionable C/E data provided by CBECS respondent | |||
---|---|---|---|---|---|
No ESS | To ESS for research purposes | To ESS | No ESS because already decided to impute | ||
Electricity | 6,258 | 2,568 | 409 | 3,276 | 5 |
Natural gas | 4,132 | 1,455 | 260 | 2,391 | 26 |
Fuel oil | 1,498 | 814 | N/A | 70 | 614 |
District heat | 407 | 174 | N/A | 36 | 197 |
Phase 2: energy suppliers survey
The ESS data collection began in March 2014, several months after the end of the CBECS buildings data collection. An ESS data collection website offered the energy suppliers multiple reporting options: an interactive data collection web page, an Excel spreadsheet that could be downloaded and completed by the supplier and submitted online, an option to create and submit their own electronic file through the website, or the option to print out the data collection forms and mail them to the survey contractor.
Suppliers were provided the addresses and account numbers (where available) for the CBECS buildings for which EIA was requesting data. The data requested was energy usage and cost information for the building for up to 16 billing periods covering November 2011 through February 2013. For buildings with multiple accounts, suppliers were given the option to provide data at the whole building level or for each account within the building.
The total number of suppliers that had to be contacted for the CBECS sample was 299 electricity, 182 natural gas, 59 fuel oil, and 12 district heat.
Unlike the CBECS buildings survey, response to the ESS is mandatory. Almost all of the suppliers provided data for at least some of their customers, but there was some building-level nonresponse, mostly in cases where suppliers could not locate the customer in their records or the building’s supplier could not be determined. The next table shows the number of cases by energy source for which ESS data had been received at the end of ESS data collection.
Cases sent to ESS | Of these, cases where ESS data were received | |
---|---|---|
Electricity | 3,685 | 3,325 (90%) |
Natural gas | 2,658 | 2,318 (87%) |
Fuel oil | 70 | 53 (76%) |
District heat | 35 | 32 (91%) |
After the ESS data were received at EIA, another round of review began. The process was similar to the checking of the CBECS data; the energy intensity by building activity was checked against previous CBECS values. Additionally, the data were checked by billing period for date gaps or overlaps, unusually long or short billing periods, prices that appeared too high or too low, and for drastic changes in usage from one billing period to another. The number of cases reviewed by energy source is shown in the table below, along with the number of cases for which the ESS data were determined to be useable.
Cases where ESS data were received | Cases reviewed by CBECS team | Cases where ESS data were useable | |
---|---|---|---|
Electricity | 3,326 | 1,951 (59%) | 3,030 (91%) |
Natural gas | 2,318 | 1,390 (60%) | 2,159 (93%) |
Fuel oil | 53 | 17 (32%) | 50 (94%) |
District heat | 32 | 15 (47%) | 28 (88%) |
At this point, because of the additional cases added to the ESS for research, there were some cases with values from both the CBECS and the ESS. The CBECS team at EIA reviewed 238 electricity and 193 natural gas cases to determine the final data source.
The final sources for the consumption and/or expenditures data at the end of all the phases of data receipt, processing, and editing are shown in the following table.
Cases using the energy source | C/E from CBECS | C/E from ESS | C/E imputed | |
---|---|---|---|---|
Electricity | 6,258 | 2,849 (46%) | 2,826 (45%) | 583 (9%) |
Natural gas | 4,117 | 1,628 (40%) | 2,011 (49%) | 478 (12%) |
Fuel oil | 1,497 | 794 (53%) | 50 (3%) | 653 (44%) |
District heat | 407 | 169 (42%) | 28 (7%) | 210 (52%) |
With the exception of fuel oil, the percent of cases with imputed consumption in the 2012 CBECS is lower than in the 2003 CBECS, which has positive implications for the quality of the 2012 CBECS data. A comparison of the percent imputed is shown in the following table.
Of the cases using the energy source, percent imputed in the 2003 CBECS | Of the cases using the energy source, percent imputed in the 2012 CBECS | |
---|---|---|
Electricity | 31% | 9% |
Natural gas | 27% | 12% |
Fuel oil | 36% | 44% |
District heat | 58% | 52% |